This paper by David Blanchflower and Andrew Oswald (from the Australian Economic Review in 2005) looks interesting. I'm interested in happiness (who isn't?) but this paper particularly interests me because it addresses a special case of the general statistical problem of summarizing multivariate data by indexes. Here's the abstract:
According to the well-being measure known as the U.N. Human Development Index, Australia now ranks 3rd in the world and higher than all other English-speaking nations. This paper questions that assessment. It reviews work on the economics of happiness, considers implications for policymakers, and explores where Australia lies in international subjective well-being rankings. Using new data on approximately 50,000 randomly sampled individuals from 35 nations, the paper shows that Australians have some of the lowest levels of job satisfaction in the world. Moreover, among the sub-sample of English-speaking nations, where a common language should help subjective measures to be reliable, Australia performs poorly on a range of happiness indicators. The paper discusses this paradox. Our purpose is not to reject HDI methods, but rather to argue that much remains to be understood in this area.
I recommend--for the next paper these folks write--to present the results in graphical, not tabular, form, and to order the countries in some reasonable way (for example, in order of per-capita GDP) rather than alphabetically. For example, do we really need to know that Australia has a value of 5.39 for one index and 5.62 for another? These comments apply to the raw data and also the displays of regression coefficients.
Going on to the substance of the paper, I have no particular comments. It is admirably crisp and speaks for itself and modestly focuses on the statistical issues.
Leave a comment